Terminal-based agentic AI coding assistant from Charm that works with many LLM providers, uses LSP and MCP for context, manages per-project sessions, and asks permission before running tools by default.
Open the source and read safety notes before installing.
Source-backed facts for citing this resource, derived directly from the registry — also available as plain text for AI assistants.
Decision playbook
Signals are present but mixed. Use the checklist below to confirm the source and operational safety for your environment.
0
63
—
No baseline selected
No major trust-signal divergence detected in the current selection.
Confirm ownership and provenance before trusting install instructions.
Source link availableRequired
Open the canonical repository and verify ownership.
Source provenance statusRequired
Marked as source-backed.
Metadata reviewed
No reviewed flag detected in metadata.
Validate risk disclosures before installation or API wiring.
Safety notes presentRequired
Review the listed safety guidance before running commands.
Privacy notes presentRequired
Review data handling notes before connecting accounts or secrets.
Trust level risk gateRequired
Trust level does not block evaluation.
Check package metadata and artifact integrity signals.
Install payload available
Install or copy payload is available for review.
Package verification flag
No package verification flag provided.
Checksum metadata
No checksum provided for downloaded artifact.
Use compare context to validate trade-offs before adoption.
Compare tray has multiple entries
Add at least one more entry to compare trust differences.
Baseline comparison available
No baseline peer selected yet.
Diverging trust signals identified
No major trust-signal divergence found.
Setup at a glance
Copy-ready — paste the snippet to get started.
Adoption plan
Current risk score 24/100. Use staged verification before broader rollout.
Validate source and review signals before any execution.
Confirm source provenanceRequired
Source URL/provenance metadata is present.
Confirm metadata review state
No review metadata found; increase manual validation.
Verify install payload
Install/config payload exists and can be inspected.
Confirm safety, privacy, and package integrity signals.
Review safety notesRequired
Safety notes are present.
Review privacy notesRequired
Privacy notes are present.
Verify package integrity metadata
No package verification/checksum metadata.
Adopt in controlled steps based on the selected plan.
Run in isolated sandbox firstRequired
Use a constrained sandbox and observe behavior across multiple tasks.
Roll out graduallyRequired
Roll out to a small cohort before wider usage.
Set monitoring and fallback
Define rollback path and monitor errors after adoption.
Evidence readiness
Missing required evidence: Metadata review. Risk score 31.
Source repository/provenance is listed.
Required in this preset
Review metadata is missing.
Required in this preset
Safety notes are present.
Required in this preset
Privacy notes are present.
Optional in this preset
Package integrity metadata is missing.
Optional in this preset
Install payload is available.
Required in this preset
Required gaps: Metadata review
Decision timeline
Blocking gaps: Check metadata review status. Risk 28.
triage
Source/provenance metadata is available.
triage
Review metadata is missing.
verify
Safety notes are available.
verify
Privacy notes are available.
verify
Package integrity metadata is missing.
rollout
Install payload is available.
Blockers: Check metadata review status
Prerequisite readiness
3 prerequisites to line up before setup. Have accounts and credentials ready first.
Safety & privacy surface
4 safety and 3 privacy notes across 6 risk areas. Review closely: credentials & tokens, permissions & scopes, third-party handling.
Disclosure: editorial
# Homebrew
brew install charmbracelet/tap/crush
# npm
npm install -g @charmland/crush
# Go
go install github.com/charmbracelet/crush@latestCrush is a terminal-based agentic AI coding assistant from Charm — "glamourous agentic coding for all." It bridges your terminal, your code, and the LLM of your choice, analyzing your codebase, using LSP for context, and collaborating on coding tasks directly in the terminal.
It is written in Go and released under the Functional Source License (FSL-1.1-MIT): the source is public and free to use, and converts to MIT over time. You bring your own LLM provider API key.
# Homebrew
brew install charmbracelet/tap/crush
# npm
npm install -g @charmland/crush
# Go
go install github.com/charmbracelet/crush@latest
Editorial listing. No paid placement or affiliate relationship. Crush is source-available under FSL-1.1-MIT; using it requires your own paid or free LLM provider access.
Show that Crush is listed on HeyClaude. Paste this Markdown into your README — it renders the badge and links back to this page.
[](https://heyclau.de/entry/tools/crush)Crush side by side with 3 alternatives on trust, install, platform support, and disclosed safety notes — all from reviewed registry metadata.
1 trust signal differ across this comparison (Submitter).
Next steps differ across entries — use the actions in the table below to copy install commands and source links per resource.
| Field | Terminal-based agentic AI coding assistant from Charm that works with many LLM providers, uses LSP and MCP for context, manages per-project sessions, and asks permission before running tools by default. Open dossier | Lightweight, open-source coding agent from OpenAI that runs in your terminal, edits and runs code locally with configurable sandbox and approval modes, and authenticates with a ChatGPT plan or API key. Open dossier | Google's open-source terminal AI agent for Gemini-powered coding and automation, with code understanding, file edits, shell commands, web fetching, Google Search grounding, MCP server integrations, checkpointing, GEMINI.md context files, and GitHub workflow automation. Open dossier | Terminal-first AI coding agent for local development workflows, codebase edits, and model-flexible automation. Open dossier |
|---|---|---|---|---|
| Next stepsDiffers | ||||
| Trust | ||||
| Review status | Not reviewed | Not reviewed | Not reviewed | Not reviewed |
| Package trust | Package not verified | Package not verified | Package not verified | Package not verified |
| Source provenance | Source-backed | Source-backed | Source-backed | Source-backed |
| SubmitterDiffers | JPette1783 | JPette1783 | — | — |
| Install risk | Review first | Review first | Review first | Review first |
| Notes | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ |
| Brand | — | |||
| Category | tools | tools | tools | tools |
| Source | Source-backed | Source-backed | Source-backed | Source-backed |
| Author | Charm | OpenAI | SST | |
| Added | 2026-06-05 | 2026-06-05 | 2026-06-18 | 2026-04-27 |
| Platforms | ||||
| Harness | ||||
| Source repo | — | — | — | — |
| Safety notes | ✓Crush executes tools and commands; by default it asks for permission before each tool call. The --yolo flag skips all permission prompts; the project warns to be very careful with it, so avoid it on untrusted work. The crush.json config is trusted code — any $(...) in it runs at load time with your shell's privileges, so review config files before use. LSP and MCP servers can read your codebase and influence agent behavior; only connect servers you trust. | ✓Codex CLI runs as a local agent that can edit files and execute commands on your machine. It supports configurable sandbox and approval modes; keep approvals enabled and review proposed actions before allowing them. Lowering approval or sandbox restrictions lets the agent act more autonomously, increasing the risk of unintended changes. Review generated code and commands before running them against important repositories or systems. | ✓Gemini CLI can read and edit local files, run shell commands, fetch web content, use Google Search grounding, and call configured MCP servers; keep it inside version-controlled workspaces and review high-impact actions. MCP integrations can expose databases, SaaS accounts, browsers, cloud resources, files, or internal APIs to the agent; apply least privilege and approval gates per server. Preview and nightly release channels may contain regressions or unvetted changes; use stable releases for shared or production workflows unless testing intentionally. Non-interactive scripting can run without the same operator attention as an interactive session; constrain prompts, output parsing, credentials, and command permissions. GitHub workflow automation through Gemini CLI should be reviewed like any other code-review or issue-triage automation before granting repository permissions. | ✓OpenCode is an agent that reads, edits, and can run code in your local repository; review proposed changes and run it in version-controlled projects. |
| Privacy notes | ✓Your code context and prompts are transmitted to the LLM provider you configure. API keys are read from environment variables or config files and sent to the configured provider; store them as secrets. Crush records pseudonymous usage metrics tied to a device-specific hash; prompts and responses are never collected. Opt out with CRUSH_DISABLE_METRICS=1 or DO_NOT_TRACK=1. | ✓Code and prompts are sent to OpenAI to power the agent under your ChatGPT plan or API key. Data handling follows OpenAI's policies for the plan or API you authenticate with; review them before processing sensitive code. API keys and account credentials should be stored securely and never committed to source control. | ✓Prompts, selected source files, GEMINI.md context, shell output, web fetches, MCP tool arguments, MCP tool results, checkpoints, and command output may be sent through the configured Gemini or Vertex AI route. Keep API keys, Google Cloud project IDs, service credentials, private paths, customer data, and internal code out of prompts, logs, shared terminal output, and public issues. Google account, Gemini API, Vertex AI, retention, quota, telemetry, and billing behavior depend on the selected authentication mode and organizational settings. MCP server logs, Gemini CLI logs, terminal history, GitHub workflow logs, and generated artifacts can retain sensitive code or operational context. | ✓OpenCode sends your code, prompts, and file context to the configured LLM provider to plan and apply edits; choose providers deliberately and keep secrets out of shared context. |
| Prerequisites |
|
|
| — none listed |
| Install | — | — | | — |
| Config | — | — | — | — |
| Citations | ||||
| Claim | Unclaimed | Unclaimed | Unclaimed | Unclaimed |
OpenAI's open-source terminal coding agent that edits and runs code locally with sandbox and approval modes.
Use Google's Gemini CLI as a terminal AI agent for codebase work, shell commands, web fetching, Search grounding, MCP integrations, checkpointing, and scripted automation.
Source-backed guides for putting this to work.
Write a claude-code-hint tag to stderr under CLAUDECODE, and Claude Code prompts users once to install your official-marketplace plugin.
Loading live community signals…
A short, calm digest of reviewed Claude resources. Unsubscribe any time.